Hacker News Reader

Comment by loeg | Hacker News Reader

> I'm wondering how do they test it, do they inject volunteers with the vacine, and they tell them to go infect themselves with a virus that 1 time out of 100 will kill them and an unknown percentage disable them?

No, and this is why Phase III trials take a long time, potentially a very long time if we actually get a handle on the virus with measures like masking and social distancing.

Here's the protocol:

1. Volunteers get the vaccine (experimental arm) or a placebo (control arm).

2. They go about their lives.

3. Both study arms report any cases of COVID they develop. Also, any deleterious symptom (potential side effects).

4. If the vaccine arm has a statistically significant reduction in infection rates, and/or severity, everyone wins and the study is concluded.

This process can take years if the background rate of infection is low enough.

View on HN

> This process can take years if the background rate of infection is low enough.

This is why the phase 3 trials tend to be organised in countries with high rates of infection, such as Brazil, South Africa or India. In locations with low rates of infection it's too likely that the people in the experimental group don't get infected simply because they don't get exposed to the virus.

> This is why the phase 3 trials tend to be organised in countries with high rates of infection, such as Brazil, South Africa or India.

Or US (we are still higher than Brazil in cases per million). South Africa or India are actually much lower than us.

I can see why it may take years to determine if the infection rate has been significantly lowered.

But what is the timeline for determining if there are serious deleterious symptoms?

If we get to the point where we know a bunch of vaccines are safe, but not necessarily effective, could these be approved for anyone who understands the risks?

Adverse events are monitored continuously as long as the patient is still a part of the trial (which often extends past approval).

Once the product is approved, the FDA has an adverse event monitoring system that collects reports and analyzes them for signals that might imply a previously-unknown adverse event.

Of course it take more than just one case to for a new signal to be identified (unless it's really obvious like the patient dies right after getting the vaccine), so sometimes it may be a year or more until there is enough data to say "yes, we know nearly all the risks associated with this vaccine".

Sure, that makes sense.

My comment was based on Fauci's recent comments saying approval could happen as soon as next month if there are enough infections in the control group.

It seems to me that what he is implying is that the risk of long-term side effects would not be worth blocking approval of a vaccine if the vaccine proved to be effective in preventing the virus.

Gotcha. Yeah, that timeline seemed really aggressive, but I would hope that the FDA uses it's standard set of evaluating criteria for any new vaccine. Even though the phase 3 might only have a few months of data, you can pool phase 1 + 2 data as well, which would give you a bit more confidence as to potential long term side effects.

It would also depend on what they approve it for. You could see them restricting it to high-risk patients. In that case the risk-benefit is much better, so you'd be willing to take a chance on some hidden adverse event popping up since you'd be looking at avoiding a significant risk of death from Covid).

At least in my experience, there isn't a set formula for evaluating new drugs. The FDA really does look at every piece of data it can find. I've also found the FDA to be pretty conservative - if they are unsure, they lean towards not approving, or approving with significant restrictions.

The FDA has published a detailed guideline on how to handle vaccines against the SARS-CoV-2. Among these, the minimum requirement of a 50% efficacy and mandatory monitoring of the trial participants for at least one year after the trial end (which will not be when authorization requests will be made, in case it works, but later on).

Thank you for the explanation.

Reply on news.ycombinator.com